A professional manager told me that except for professional analysts in the field of industry data research, the rest of the planning personnel in the company have no background in mathematical statistics, which means that their data analysis ability is very weak. This is a 100 billion-level enterprise. Planning employees have been trained by mentoring for many years, mainly focusing on data collection and aggregation, and heavily relying on the judgment of sales, products and senior executives in demand forecasting.
When it comes to “starting from data”, many people think of mathematical models statistics. What I want to say is that those are not the same thing: in data analysis, data models and mathematical statistics only account for a small part. For planning, starting from data is not so much about mathematical statistics and data models as it is a work habit: before asking internal customers such as sales, marketing, and products, or not knowing where to start, first analyze the data, summarize regular things, identify potential problems, and then find right people to confirm, judge, and adjust. In these analysis, most tasks can be solved with simple addition, subtraction, multiplication, and division, which can be implemented in Excel tables, without knowing much mathematical statistics and models.
Seeing that she was a little confused, I gave her an example.
There was a company that had always had great challenges in preparing for holiday season. I found one of their main products and guided their cross-functional team to design a set of ” expert judgment method” based on this product to integrate the wisdom of the cross-functional team, work together, and avoid big mistakes (this method can be referred to in my book “Demand Forecasting and Inventory Planning: A Practitioner’s Perspective”). In the expert judgment method, a key task is to determine expert team. After the discussion results of several groups came out, almost without exception, the expert team listed the CEOs of various functions, the directors of major accounts, and business heads of major regions and cities. This looked like a wide net: they have more than ten major customers, four regions and twenty or thirty cities, corresponding to dozens of people. You can’t bring them all in to make judgments, right?
For this product, I analyzed the delivery history and found that in the past three months, the two major customers accounted for nearly 50%, and the remaining 30% or so belonged to online retail investors (ordering through the App) and about 20% belonged to channel customers. You will immediately find that the core judges are two key account directors, plus one channel manager and one online business manager (the channel manager is in charge of formulating policies and planning channel activities, so he can make a certain judgment on holiday season stocking. The online business is similar and is managed by the online business manager).
The situation of key customers is more complicated, because those key customers are big one, and their business spread all over the country. The connection between the headquarters and the branches is not necessarily close. The key account director corresponds to the customer headquarters, and it is difficult to make judgments on specific city branches. So, we included the account managers of these three cities into the expert team, and a more targeted expert team was established.
I simply analyzed the data of the company in the first half of the year, and I could see clear fluctuations in key customers, channels and online business. In just a few weeks, the shipment volume has increased several times. There must be something happening behind this that can significantly change the demand. Some are externally driven, such as the activities of major customers themselves; some are internally driven, such as channel policies, online activities, etc. In the historical data, a little analysis, such as summarizing the shipment volume by week and making a simple line chart, can identify these major changes, and then find the right account manager, channel and online business manager to predict whether similar things will happen in the future, so that more accurate predictions can be made.
These are all based on data analysis, but there is no need for any mathematical statistics; all that is needed is the habit of looking at data.
Interestingly, in this case, I attached the above analysis results to the materials for the group discussion, but most of the groups did not even look at them, just slapped their heads there and cast a wide net based on experience. Without data support, there is a lack of targeting; looking for a needle in a haystack wastes too many resources; resources are too scattered, and there are not enough resources invested in what really needs to be focused.
No one knows more than data. I first realized this from a financial director. This director was new to the planning department and was a layman in planning. But it doesn’t matter. He locked himself in the office and fiddled with the computer for a few days. He sent out one report after another, telling the planning team that these products seemed to have this problem and those products had that problem, and asked the planners to adjust the plan. I saw that most of the “problem” products I was familiar with were on the list. In addition, there were some products that we didn’t know about but whose trends were getting worse. This made me realize the power of data. Yes, for those more than 10,000 SKUs, no one’s experience can compare with data analysis, and it’s better to ask data than anyone else.
Planning is an analytical position, and we must change the habit of relying on feedback from the business side and reacting passively. You must know that for what has happened, the business side’s knowledge is mostly partial and lagging; data analysis can often provide more timely and comprehensive information. Take the shortages and surpluses we often face as an example. How did sales find the problem? It is often because there is too much or too little inventory, and someone reported it to the sales side. But it’s too late: if you monitor the data and analyze the customer’s order and delivery history, you will often find clues to shortages or surpluses a few weeks in advance. In fact, none of these require much data analysis ability, but rather a work habit of “starting from data.”